CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition SYNTACTIC ERGATIVITY Maria Polinsky Harvard University Abstract This chapter presents and analyzes the phenomenon of syntactic ergativity, defined as the grouping of the absolutive subject and absolutive object into a natural class, to the exclusion of the ergative argument, with respect to A’-movement. Presently, there is no consensus in the literature as to the cause of syntactic ergativity. Two families of approaches can be distinguished: those which place the explanatory burden on the derivation of the absolutive, and those which invoke the properties of the ergative expression itself to explain syntactic ergativity. For the first family of approaches, which include explanations based on locality and A’-movement for case, the exclusion of the ergative from A’-movement is simply a side effect of satisfying Case licensing needs of the absolutive. In the second type of approach, the restriction on A’-movement of the ergative follows either from criterial freezing or from the adpositional-phrase nature of the ergative expression. The approaches converge on the notion that ergative is an inherent case, assigned either directly by a verbal head or by an adposition selected by a v head. Keywords: A’-movement, absolutive, ergative, syntactic ergativity 1 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition 1 Introduction 1.1 Morphological and syntactic ergativity Linguists have long classified languages according to the ways in which their intransitive subjects, transitive subjects, and direct objects align with respect to case marking and/or agreement. The two main divisions are known as the (nominative-)accusative and ergative(-absolutive) alignments. Under an accusative alignment pattern, the intransitive subject (abbreviated here as S) and the transitive subject (A: for agent, or agent-like argument) are encoded the same way (nominative), while the transitive direct object (O) is encoded separately (accusative). Under an ergative alignment pattern, on the other hand, S and O have identical encoding (absolutive) while A has its own separate case (ergative); see Comrie (1978), Dixon (1979, 1994), Manning (1996), Aldridge (2008), McGregor (2009), a.o. These alignments can be expressed not only through case marking but also through agreement; S and A may determine the same agreement, in contrast to O, or S and O may license the same agreement, in contrast to A. The two alignment patterns are illustrated in (1): (1) a. Accusative b. Ergative S S A O A O 2 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition Examples (2) and (3) illustrate accusative and ergative alignment as expressed by case marking, otherwise known as dependent marking (see Nichols 1986 on the opposition between dependent-marking and head-marking languages).1 (2) a. Kodomo-tati-ga Tokyo-ni sun-de i-ta. Japanese child-PL-NOM.[S] Tokyo-LOC live-GER be-PST ‘The children lived in Tokyo.’ b. Sensei-tati-ga nihongo-o hanasi-ta. teacher-PL-NOM.[A] Japanese-ACC.[O] speak-PST ‘The teachers spoke Japanese.’ (3) a. Kua nofo e tau fānau i Niue. Niuean PFV live [ABS PL children].[S] in Niue ‘The children lived in Niue.’ b. Kua fakaaoga he tau faiaoga e vagahau niue. PFV speak [ERG PL teacher].[A] [ABS language Niue].[O] ‘The teachers spoke the Niuean language.’ Examples (4) and (5) illustrate the expression of alignment through verbal agreement, i.e., head marking. Note that in these latter examples, the DPs appear without any overt case markers. In the Palauan examples, both S and A are indexed by the marker ng on the verb while agreement with the O argument is different (-ii); hence, the pattern is 1 Unless indicated otherwise, the examples are from the author’s fieldnotes. 3 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition accusative. In the Abkhaz examples, the S and O arguments determine absolutive agreement on the verb, and the agreement with the A is different, which leads to the ergative pattern. (4) a. Ng merael a chais er a beluu. Palauan 3SG.[S] go DET news PRP DET area ‘A rumor is going around.’ (Nuger 2010, 45) b. Ng mo kol-ii a bobai pro. 3SG.[A] AUX.FUT eat.PF-3SG.[O] DET papaya ‘He is going to eat (up) the papaya.’ (Nuger 2010, 87) (5) a. A-ts′gwǝ ∅-psǝ-ø-jt′. Abkhaz DET-cat 3SG.ABS.[S]-die-PST-FIN ‘The cat died.’ (Hewitt 2005, 39) w j b. A-ab′dǝw a-� ǝ′tʃ′ǝ a-′� ǝ DET-grandfather DET-child DET-gold ∅-′jǝ-j-ta-∅-jt′. 3SG.ABS.[O]-3SG.IO-3SG.ERG.[A]-give-PST-FIN ‘Grandfather gave the child (the) gold.’ (Keenan 2013, 5) The expression of ergative alignment through case marking and/or agreement is known as morphological ergativity, because of the visible morphological realization. However, the contrast between accusative and ergative alignment can also be found beyond morphology, in the differential behavior of S and O, on the one hand, and A, on the other, 4 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition with respect to various syntactic phenomena. This is the notion of syntactic ergativity, a preliminary definition for which is given below: (6) SYNTACTIC ERGATIVITY: TAKE 1 The presence of syntactic rules that group S and O (the absolutive) together, to the exclusion of A (the ergative). The basic pattern of syntactic ergativity is illustrated below for the Polynesian language Tongan. In Tongan, the absolutive DP can relativize leaving a gap at the extraction site, but relativization of the ergative DP requires that a resumptive pronoun ne appear in the relative clause. Without that resumptive pronoun, the relativization of the ergative DP is impossible. Consider the following examples:2 (7) a. Baseline intransitive sentence Tongan ‘Oku malimali ‘a e ta’ahine. PRS smile ABS DET girl ‘The girl is smiling.’ b. relativization of the absolutive subject with a gap ‘a e ta’ahinei [‘oku malimali __i ] 2 Here and below the gap at the extraction site is represented atheoretically as ___ with a subscript. 5 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition ABS DET girl PRS smile ‘the girl who is smiling’ (8) a. Baseline transitive sentence Tongan ‘Oku ‘ene ‘e he tamasi’i ‘a e ta’ahine. PRS tickle ERG DET boy ABS DET girl ‘The boy is tickling the girl.’ b. relativization of the absolutive object with a gap ‘a e ta’ahinei [‘oku ‘ene ‘e he tamasi’i __i ] ABS DET girl PRS tickle ERG DET boy ‘the girl whom the boy is tickling’ c. relativization of the ergative subject with a resumptive pronoun *‘a e tamasi’ii [‘oku ‘ene ‘a e ta’ahine] ABS DET boy PRS tickle ABS DET girl ‘the boy who is tickling the girl’ ‘a e tamasi’ii [‘oku nei ‘ene ‘a e ta’ahine] ABS DET boy PRS RP tickle ABS DET girl ‘the boy who is tickling the girl’ Contrast this pattern with the pattern of extraction observed in Basque, where both the ergative argument and the absolutive argument can leave a gap at the extraction site (Hualde and Ortiz de Urbina 2003, 774). Thus, all three core arguments undergo A’- movement in the same manner, leaving a gap at the extraction site: 6 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition (9) a. baseline intransitive sentence Basque Haur guzti-a-k eskapa-tzen dira. children all-DET-PL escape-IPFV AUX ‘All the children run away.’ b. ABS subject extracts with a gap at the extraction site [__i eskapa-tzen dir-en] haur guzti-a-k escape-IPFV AUX-ADN children all-DET-PL ‘all the chidren that run away’ (10) a. baseline transitive sentence Mutiko-a-k pinguinu-a-Ø garbi-tzen du. boy-DET-ERG penguin-DET-ABS wash-IPFV AUX ‘The boy is washing the penguin.’ b. ABS object extracts with a gap at the extraction site [mutiko-a-k ___ i garbi-tzen du-en] pinguinu-ai boy-DET-ERG wash-IPFV AUX-ADN penguin-DET ‘the penguin that the boy is washing’ c. ERG subject extracts with a gap at the extraction site [___ i pinguinu-a-Ø garbi-tzen du-en] mutiko-ai penguin-DET-ABS wash-IPFV AUX-ADN boy-DET ‘the boy that is washing the penguin’ (Gutierrez-Mangado and Ezeizabarrena 2012) 7 CH 105: Syntactic Ergativity, to appear in The Companion to Syntax, 2nd Edition 1.2 Setting the boundaries Several questions arise at this point, including a number that are beyond the scope of this chapter. The first question has to do with the relationship between morphological and syntactic ergativity: is the expression of syntactic ergativity limited to morphologically ergative languages? To anticipate the discussion below, the answer is yes. There are morphologically ergative languages that do not show any syntactic ergativity, but there seem to be no instances of syntactic ergativity beyond the realm of morphologically ergative languages. A second question has to do with the extent to which a language embraces syntactic ergativity: can a language be completely syntactically ergative or completely syntactically accusative? Researchers have shown that the postulation of a global contrast between syntactically accusative and syntactically ergative languages is too simplistic (see especially Kazenin 1994 for criticism of such a contrast). However, there are certain syntactic phenomena that repeatedly group S and O together to the exclusion of A, and thus deserve consideration. Such phenomena and their explanations are the focus of this chapter. Explaining syntactic ergativity also requires us to seek an answer to a third question: can we predict whether a language will be syntactically ergative or syntactically accusative? At present, we are not in a position to make such predictions; however, existing accounts of syntactic ergativity have established some correlations that may bring us closer to a satisfying predictive account.